Design method of marine equipment resilience manufacturing system based on bayesian algorithm

Abstract

This invention discloses a design method for resilient manufacturing systems of marine engineering equipment based on Bayesian algorithms. The method includes defining system productivity as a quantitative index of system resilience; constructing a discrete event simulation model of the impact of key parameters such as equipment layout, material transport time, and equipment configuration quantity on system productivity; designing a multi-source disturbance injection mechanism; generating system resilience curves and calculating resilience values ​​through the discrete event simulation model; setting the range of optimization variables based on the discrete event simulation model; employing a Bayesian optimization algorithm based on a Gaussian process surrogate model to perform global iterative optimization with the goal of maximizing system resilience; using the Sobol global sensitivity analysis method to quantify the influence weight of each parameter on system resilience and identify key bottlenecks in the manufacturing system; and outputting the optimal equipment configuration and spatial layout static parameters. This invention effectively improves parameter optimization efficiency, reduces solution search costs, and enhances the consistency between simulation results and actual production processes.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and manufacturing system optimization technology for marine engineering equipment, and in particular to a design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithms. Specifically, it relates to a manufacturing system design method based on discrete event simulation, multi-source disturbance modeling, resilience quantification evaluation, and Bayesian optimization. Background Technology

[0002] As a crucial component of marine resource development and utilization, marine engineering equipment typically involves multiple processes such as steel pretreatment, cutting, welding, assembly, pre-assembly, and final assembly. This process is characterized by long workflows, diverse equipment types, complex logistics and transportation, and high production organization difficulty. The operational efficiency and stability of the manufacturing system directly impact the manufacturing cycle, production cost, and delivery capability of marine engineering equipment products. Therefore, manufacturing system design and optimization have become important research topics in the field of intelligent manufacturing for marine engineering equipment.

[0003] Currently, the design of marine engineering equipment manufacturing systems mainly revolves around equipment configuration, equipment layout, and logistics transportation schemes. During the design process, the number of equipment and spatial layout are typically determined based on historical production experience, process requirements, and equipment utilization indicators, and production efficiency is verified through discrete event simulations. This method can meet production needs to a certain extent, but its optimization objectives are mainly focused on traditional indicators such as production efficiency, equipment utilization, and production costs. It fails to adequately consider random disturbances such as equipment failures, logistics delays, and production fluctuations, making it difficult to comprehensively reflect the stable operation capability of the manufacturing system under complex operating environments.

[0004] Meanwhile, existing manufacturing system optimization methods mostly employ empirical adjustments, parameter experiments, or traditional heuristic optimization algorithms for system design. Due to the complex coupling relationships between the number of equipment, equipment layout, and logistics transportation parameters, the system design space exhibits high-dimensional and nonlinear characteristics, making it difficult for traditional optimization methods to quickly obtain the global optimum under limited computational resources. Furthermore, the extensive simulation calculations of numerous discrete events lead to long optimization cycles, further limiting the engineering application efficiency of manufacturing system parameter optimization methods.

[0005] Therefore, although existing methods can achieve manufacturing system modeling and performance analysis, the following problems still exist: First, there is a lack of a quantitative evaluation mechanism with system resilience as the core objective; second, there is a lack of methods for joint modeling of multi-source disturbances and dynamic propagation analysis; third, it is difficult to establish a quantitative relationship between equipment configuration, equipment layout, logistics and transportation parameters and system resilience; and fourth, the optimization efficiency is low and the ability to identify key influencing factors is lacking. These problems limit the further improvement of the resilience design and optimization level of marine engineering equipment manufacturing systems. There is an urgent need to propose a design method that can achieve quantitative assessment of resilience, joint modeling of multi-source disturbances, and collaborative optimization of system parameters to improve the production stability and recovery capability of marine engineering equipment manufacturing systems under complex disturbance environments. Summary of the Invention

[0006] To overcome the aforementioned problems in the existing technology, this invention proposes a design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithms.

[0007] The technical solution adopted by this invention to solve its technical problem is: a design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm, comprising the following steps; Step 1: Define the system productivity as a quantitative index of the resilience of the marine engineering equipment manufacturing system, construct a discrete event simulation model of the impact of key parameters such as equipment layout, material transportation time and equipment configuration quantity on system productivity, and complete the production line model construction. Step 2: Design a multi-source disturbance injection mechanism, generate system toughness curves and calculate toughness values ​​using the discrete event simulation model constructed in Step 1; Step 3: Based on the discrete event simulation model constructed in Step 1, set the range of optimization variables, and use the Bayesian optimization algorithm based on the Gaussian process surrogate model to carry out global iterative optimization solution with the goal of maximizing system resilience; Step 4: Using the Sobol global sensitivity analysis method, the influence weight of each parameter on the system resilience is quantified, and the key bottleneck links of the manufacturing system are identified. Step 5: Output the optimal device configuration and spatial layout static parameters.

[0008] The above-mentioned design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithms, specifically step 1, is as follows: Step 1.1: Based on the actual processing scenario of marine engineering equipment, define the equipment layout matrix, configure processing stations, logistics and transportation equipment and buffer storage areas, and complete the model building; Step 1.2: After completing the modeling of equipment processing capacity, equipment layout, and material transportation time, a system productivity calculation model is established to quantify the production capacity of the manufacturing system per unit time. Productivity under theoretical conditions For workstation Production time; Step 1.3: Run the discrete event simulation model, dynamically adjust equipment fault parameters, maintenance sequence and system control rules, and iteratively correct and refine the details of the manufacturing system operation process.

[0009] The aforementioned design method for resilient manufacturing systems of marine engineering equipment based on Bayesian algorithms, specifically defining the equipment layout matrix in step 1.1, is as follows: Equipment Layout Matrix ; The equipment layout matrix must satisfy the following constraint: each piece of equipment can only be placed in one location: And at most one device can be placed at each location: .

[0010] The above-mentioned design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithms, specifically step 2, is as follows: Step 2.1: Establish a multi-source disturbance mechanism that includes equipment failure disturbances and material delay disturbances, and define their respective probability distributions; Step 2.2: Define a joint perturbation injection mechanism, randomly generate event sequences according to their respective distributions on the simulation time axis, and introduce a system perturbation coupling function; Step 2.3: Run a discrete event simulation with perturbation, record the change of the instantaneous output rate Q(t) of the system over time, and plot the time-output rate curve, i.e. the resilience curve; Step 2.4: After obtaining the system time-production rate resilience curve, in order to quantitatively characterize the comprehensive performance of the marine engineering equipment manufacturing system in maintaining and restoring production capacity under disturbance conditions, the system resilience index is calculated based on the system dynamic production rate curve.

[0011] The aforementioned design method for resilient manufacturing systems of marine engineering equipment based on Bayesian algorithms, specifically step 2.1, involves: Define the equipment fault disturbance, and based on the probability density function of the fault occurrence, obtain the system reliability function as follows: ,in, The failure rate of device i; Define the material delay disturbance, and the actual system transportation time is: ,in, Indicates time The additional delay time generated follows a random probability distribution: P(θ) is the delay probability distribution function; This indicates the system's average transit time.

[0012] The above-described design method for a robust manufacturing system for marine engineering equipment based on Bayesian algorithms, specifically the system perturbation coupling function in step 2.2, is as follows: in, The impact of equipment failure on strength, For the overall system latency intensity, To prevent the impact of maintenance on strength, Here, represents the weighting influence coefficient; 𝑚 represents the total number of devices. For time The number of devices that fail at any given time. For the processing capacity of equipment A; This represents the total number of transportation tasks currently in the system. For the first The delay time of each transportation task, This represents the theoretical transportation time for the corresponding task; For a moment The number of devices currently under maintenance. For equipment Maintenance duration, This represents the current simulation runtime.

[0013] The above-described design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithms, specifically the instantaneous output rate Q(t) of the system in step 2.3, is as follows: in, Productivity under theoretical conditions For workstation Production time.

[0014] The above-described design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithms, in step 2.4, defines the system resilience index as follows: .

[0015] The above-mentioned design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm, specifically step 4, involves: using the Sobol global sensitivity analysis method to perform global sensitivity analysis on equipment layout parameters, equipment quantity parameters, and logistics transportation time parameters; and identifying key parameters and key workstations that have a significant impact on system resilience by calculating the first-order sensitivity index and total sensitivity index of each parameter.

[0016] The above-mentioned design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm, specifically step 5, involves: combining the Bayesian optimization results and the Sobol global sensitivity analysis results to comprehensively evaluate the system parameters, select the optimal parameter combination that meets the engineering constraints, and output the equipment configuration scheme, equipment spatial layout scheme, and logistics transportation time parameters that maximize system resilience.

[0017] The beneficial effects of this invention are: (1) It constructs a resilience quantitative evaluation model with system productivity as the core, generates system resilience curves through discrete event simulation, and quantitatively characterizes the process of production capacity decline and recovery under disturbance. Compared with traditional methods that rely on experience judgment or single performance index evaluation, it can more accurately reflect the disturbance resistance and recovery capability of the manufacturing system, and provide a unified quantitative basis for the resilience design and optimization of the manufacturing system.

[0018] (2) By establishing a joint modeling mechanism for multi-source disturbances such as equipment failure, maintenance activities, and material delays, and considering the propagation and coupling relationships between disturbances, the operating state changes of manufacturing systems under complex production environments can be simulated more realistically. This method improves the consistency between simulation results and actual production processes, and provides reliable support for system resilience assessment and disturbance rejection capability analysis.

[0019] (3) Equipment configuration parameters, equipment layout parameters and logistics transportation parameters are uniformly incorporated into the manufacturing system modeling and analysis framework. By establishing a quantitative relationship between key parameters and system resilience, the system analysis of the influence of each design parameter on system performance is realized, providing a theoretical basis for manufacturing system design optimization and scheme decision-making.

[0020] (4) Combining Bayesian optimization algorithm and Sobol global sensitivity analysis method, key influencing factors are identified while achieving global optimization of the system. This method can effectively improve parameter optimization efficiency, reduce solution search cost, and obtain the equipment configuration scheme, equipment layout scheme and logistics transportation parameters with the greatest system resilience, providing technical support for the optimized design of marine engineering equipment manufacturing systems. Attached Figure Description

[0021] Figure 1 This is a simplified flowchart of the present invention; Figure 2 This is a toughness curve; Figure 3 This is a flowchart illustrating the overall design of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0023] This invention discloses a design method for a tough manufacturing system for marine engineering equipment based on Bayesian algorithms, such as... Figure 1 As shown, it includes the following steps; Step 1: Define the system productivity as a quantitative index of the resilience of the marine engineering equipment manufacturing system, construct a discrete event simulation model of the impact of key parameters such as equipment layout, material transportation time and equipment configuration quantity on system productivity, and complete the production line model construction. Step 2: Design a multi-source disturbance injection mechanism, generate system toughness curves and calculate toughness values ​​using the discrete event simulation model constructed in Step 1; Step 3: Based on the discrete event simulation model constructed in Step 1, set the range of optimization variables, and use the Bayesian optimization algorithm based on the Gaussian process surrogate model to carry out global iterative optimization solution with the goal of maximizing system resilience; Step 4: Using the Sobol global sensitivity analysis method, the influence weight of each parameter on the system resilience is quantified, and the key bottleneck links of the manufacturing system are identified. Step 5: Output the optimal device configuration and spatial layout static parameters.

[0024] This embodiment takes a marine engineering equipment manufacturing production line as the research object and establishes a discrete event simulation model based on the FlexSim platform. The system includes 21 processing equipment and 4 temporary storage buffers. The production process includes steel pretreatment, cutting, shearing, arc welding, submerged arc welding, assembly, welding, pre-assembly and final assembly.

[0025] The system's processing equipment includes: 1 steel pretreatment unit, 2 cutting units, 2 shearing units, 2 electric arc welding units, 3 submerged arc welding units, 3 assembly platforms, 3 welding units, 2 pre-assembly platforms, 2 final assembly units, and 1 docking station. Four temporary buffer zones (not occupying workstation positions) are set up between key processes to alleviate system congestion and workpiece waiting issues. The specific process is as follows... Figure 3 As shown.

[0026] In step S1, a discrete event simulation model is constructed and the production line model is built, including: S101, based on the actual processing scenarios of marine engineering equipment, configure processing stations, logistics and transportation equipment and buffer storage areas to complete the model building; S1011: Equipment Layout Matrix Definition In this embodiment, the system has a total of 21 processing devices, and the manufacturing system has a total of 25 ( Candidate workstation locations for equipment placement.

[0027] To ensure a rational equipment layout, the equipment arrangement matrix must meet the following constraints: Each piece of equipment can only be placed in one location: And at most one device can be placed at each location: .

[0028] In this embodiment, according to the marine engineering equipment manufacturing process, the equipment for steel pretreatment, cutting, shearing, arc welding, submerged arc welding, assembly, welding, pre-assembly and final assembly are arranged in spatial layout according to the process sequence, and temporary buffer zones are set between key workstations to reduce the probability of system blockage and improve the continuity of system logistics.

[0029] S102 sets the operating parameters for each processing station, including processing speed, processing sequence, equipment layout, and transport time. The specific steps are as follows: S1021: Processing rate setting.

[0030] Each workstation Processing rate Defined as the number of workpieces that can be completed at this workstation per unit of time. Let the workstation... Include A number of parallel processing units of the same type are connected, and the rated processing time of a single unit is [missing information]. (Unit: hours / piece), then the processing speed of a single machine is: The total processing speed of this workstation is: In this embodiment, to describe the randomness of the manufacturing system's operation, the processing time of each piece of equipment is modeled using a triangular distribution. For example, the processing time of the steel pretreatment equipment, the cutting equipment, and the welding equipment are defined as follows: .

[0031] S1022: Define the processing sequence.

[0032] Define the logistics relationship matrix This indicates the flow direction of the workpiece between each station: In this embodiment, the workpiece is processed according to the process sequence of steel pretreatment → cutting → shearing → welding → assembly → pre-assembly → final assembly, so A can be written in matrix form: .

[0033] S1023 defines the equipment layout and transportation time.

[0034] In the workshop layout design, there are 25 possible locations for equipment, each corresponding to a specific two-dimensional coordinate. Arranging these coordinates row-wise yields a location matrix P, with dimensions 25×2, as follows: Equipment layout matrix is (Dimension 21×25, total number of devices 21), then the actual device location matrix L can be obtained through matrix multiplication: Each row of matrix L corresponds to the location coordinates of a device. This allows us to define the devices. With equipment Euclidean distance between them: Let the material transport speed be... (Unit: distance / time), then the transportation time Equals the transport distance divided by the transport speed: The transportation time matrix is ​​obtained by summing the transportation times between all equipment pairs. : The system's logistics transportation uniformly utilizes AGVs (Automated Guided Vehicles) for workpiece transport. Considering the safety and stability during the transport of large components in the marine engineering equipment manufacturing workshop, the average transport speed of the AGVs is set as follows: Based on the layout and scale of the workshop equipment, the average transportation distance between equipment is set as follows: The average transportation time of the system can then be expressed as: .

[0035] After completing the modeling of equipment processing capacity, equipment layout, and material transportation time, S103 establishes a system productivity calculation model to quantify the production capacity of the manufacturing system per unit time.

[0036] Set up manufacturing system workstations The above contains If multiple machines are connected in parallel, then the total processing speed of station i is: Considering the impact of workpiece transportation time between workstations, the overall production time of workstation i is defined as: Since the overall production capacity of the manufacturing system is limited by the bottleneck workstation, the theoretical productivity of the system under undisturbed conditions is defined as follows: Productivity under theoretical conditions For workstation Production time.

[0037] S104 runs discrete event simulation, dynamically adjusts equipment fault parameters, maintenance sequence and system control rules, and iteratively corrects and refines the details of the manufacturing system operation process.

[0038] Step 2: Add a multi-source disturbance mechanism, generate system toughness curves and calculate toughness values ​​through discrete event simulation.

[0039] S201, Establish a multi-source disturbance model: including equipment failure disturbances and material delay disturbances, and define their respective probability distributions (failure rates). Repair rate Preventive maintenance cycle Time Material delay probability distribution, etc.

[0040] S2011 defines equipment fault disturbances.

[0041] In actual production, equipment inevitably experiences random failures, affecting the stable output capability of the manufacturing system. To characterize this characteristic, reliability modeling of the equipment at each processing station is necessary. Let the failure rate (failure rate) of equipment i be... The unit is usually "times / hour". In engineering practice, the time of equipment failure often follows an exponential distribution, and its probability density function is expressed as: This represents the probability density of a device failing within the time interval [0, t]. Based on the probability density function of the failure, the reliability function of the system can be obtained, expressed as: in The larger the size, the lower the equipment reliability.

[0042] For a manufacturing system containing 21 pieces of equipment, the reliability functions of each piece of equipment can be arranged along the diagonal to form a 21×21 order reliability matrix. : S012, defines material delay disturbance During the operation of marine engineering equipment manufacturing systems, the transportation of materials between workstations may be affected by factors such as AGV scheduling conflicts, logistics path congestion, buffer zone blockage, and upstream production fluctuations. This can cause materials to fail to arrive at the target workstation at the scheduled time, resulting in waiting at downstream workstations, fluctuations in production line cycle time, and a decrease in system output capacity. The above phenomenon is defined as material delay disturbance.

[0043] Let the theoretical transportation time between equipment i and equipment j be: in Indicates equipment With equipment The transportation distance between them For logistics and transportation speed.

[0044] The system's logistics transportation uniformly utilizes AGVs (Automated Guided Vehicles) for workpiece transport. Considering the safety and stability during the transport of large components in the marine engineering equipment manufacturing workshop, the average transport speed of the AGVs is set as follows: Based on the layout and scale of the workshop equipment, the average transportation distance between equipment is set as follows: The average transportation time of the system can then be expressed as: AGVs perform point-to-point transportation tasks according to the workpiece processing path. When downstream equipment or temporary storage areas are busy, the AGVs need to enter a waiting queue, resulting in logistics transportation delays. Considering random factors such as AGV path congestion, buffer waiting, and scheduling conflicts, the actual system transportation time is further expressed as: in, Indicates time The additional delays caused by logistics congestion, scheduling conflicts, etc., follow a random probability distribution: ,in This is the delay probability distribution function.

[0045] S202 defines a joint perturbation injection mechanism: randomly generate event sequences according to their respective distributions on the simulation time axis, and consider the coupling between perturbations (such as maintenance triggered after a fault, and maintenance leading to downstream starvation).

[0046] To simulate the dynamic evolution of marine engineering equipment manufacturing systems under complex operating environments, a joint disturbance injection mechanism is introduced on the discrete event simulation timeline. Based on the probability distribution characteristics of equipment failure disturbances, maintenance disturbances, and material delay disturbances, a sequence of disturbance events is randomly generated and dynamically injected into the system operation process according to the event occurrence time, in order to simulate the state change process of the manufacturing system under the action of random disturbances.

[0047] Let the set of disturbance events of the system at time t be: Indicates a device malfunction event; Indicates an equipment maintenance event; This indicates a material delay event.

[0048] The equipment failure time interval is randomly generated based on the failure rate, and its occurrence process follows an exponential process. Therefore, the failure time interval of equipment 𝑖 satisfies: in, For the first Intermittent failure time for each piece of equipment. For example, the interval between failures for steel pretreatment equipment, cutting equipment, and welding equipment are defined as follows: When equipment malfunctions, the system automatically triggers a maintenance event. Equipment maintenance time is modeled using a triangular distribution. in, For the first Repair time for each piece of equipment These represent the maximum, minimum, and possible values ​​of maintenance time (in hours). For example, the maintenance time for steel pretreatment equipment, cutting equipment, and welding equipment are defined as follows: During maintenance, the corresponding workstation stops processing, potentially causing downstream workstations to enter a starvation state due to material interruption. To describe the propagation and coupling relationships between disturbances, a system disturbance coupling function is introduced: The impact of equipment failure on strength, Overall delay strength of the bit system To prevent the impact of maintenance on strength, This is the weighting influence coefficient.

[0049] Among them, the impact intensity of equipment failure It is used to characterize the degree to which a faulty device weakens the overall processing capability of the system, and is defined as: Where 𝑚 represents the total number of devices. For time The number of devices that fail at any given time. The processing capacity of equipment A. The larger the value, the more severe the loss of processing capacity due to equipment failure.

[0050] Let be the overall system delay intensity at time t, used to characterize the comprehensive impact of material transport delay on the operating state of the manufacturing system, and it is defined as follows: This represents the total number of transportation tasks currently in the system. For the first The delay time of each transportation task, This represents the theoretical transportation time for the corresponding task. The larger the value, the higher the overall logistics delay of the system, and the more significant the impact on production time and system output.

[0051] The maintenance impact intensity M(t) is used to characterize the degree to which equipment maintenance activities occupy system uptime, and it is defined as: For a moment The number of devices currently under maintenance. For equipment Maintenance duration, This represents the current simulation runtime. The larger the value, the more significant the impact of system maintenance activities on production operations.

[0052] After considering the effects of disturbance coupling, the dynamic production rate of the system is expressed as: S203 runs a discrete event simulation with perturbations, recording the system's instantaneous output rate. Plot the time-output ratio curve, i.e., the resilience curve, over time, as shown below. Figure 2 As shown: The theoretical production rate of the system under undisturbed conditions is defined as: Productivity under theoretical conditions For workstation Production time.

[0053] Considering the impact of equipment failure Overall system latency and the intensity of the impact of maintenance After the coupling effect, the overall system disturbance strength is defined as: in , This is the weighting influence coefficient.

[0054] Under the influence of disturbances, the instantaneous production rate of the system changes dynamically, and its expression is as follows: For a moment The actual production rate of the system. This represents the theoretical production rate of the system under undisturbed conditions. The overall disturbance intensity of the system.

[0055] During the discrete event simulation, a fixed time step is followed. Update the system state and record the instantaneous production rate sequence of the system at each moment: In time The horizontal axis represents the system's instantaneous production rate. Plot the system time-production rate change curve, i.e., the system resilience curve, with the vertical axis as the ordinate.

[0056] When the system is disturbed, the resilience curve will decrease; as maintenance activities are carried out and the system condition is restored, the system production rate will gradually recover, thus forming a dynamic evolution process of system performance degradation and recovery.

[0057] By analyzing the decline in production rate, recovery speed, and steady-state recovery level in the resilience curve, the resilience level of marine engineering equipment manufacturing systems under complex disturbance environments can be further quantified.

[0058] S204. After obtaining the system time-production rate resilience curve, in order to quantitatively characterize the comprehensive performance of the marine engineering equipment manufacturing system in maintaining and restoring production capacity under disturbance conditions, the system resilience index is calculated based on the system dynamic production rate curve.

[0059] Assume the theoretical production rate of the system under undisturbed conditions is... under disturbance conditions at any time The actual production rate is The normalized production performance function of the system is then defined as: When the system is in a stable, undisturbed operating state When the system is affected by equipment failure, logistical delays, or maintenance disruptions, the system's production rate decreases, leading to... .

[0060] Based on the resilience curve, the system resilience index is defined as the ratio of the area under the normalized production performance curve to the ideal operating area within the simulation period, and its expression is: As a system resilience indicator, The total system simulation time. The dynamic production rate of the system under disturbance conditions. The theoretical production rate of the system under undisturbed conditions. To normalize the production capacity of the system.

[0061] because: Therefore, the system resilience index can be further expressed as: This model shows that system resilience essentially reflects the overall level of maintaining and restoring production capacity throughout the entire operating cycle. When the system is disturbed, if the production rate drops only slightly and recovers quickly, the area under the resilience curve is large, corresponding to a higher system resilience index; conversely, if the system remains in a low-production state for a long period, the system resilience index decreases.

[0062] The above methods can be used to quantitatively calculate the resilience of marine engineering equipment manufacturing systems, providing an evaluation basis for subsequent equipment configuration optimization, equipment layout optimization, and system resilience improvement.

[0063] Step S3 specifically involves: In order to improve the production stability and recovery capability of the marine engineering equipment manufacturing system under complex disturbance environments, after completing the system disturbance modeling and resilience quantification, the system equipment layout parameters and equipment configuration parameters are jointly optimized, and a Bayesian optimization algorithm based on the Gaussian process surrogate model is used to achieve global iterative optimization of system resilience.

[0064] First, a set of design variables is constructed based on the spatial layout relationship of equipment in the manufacturing system. Among them, the equipment layout parameters are used to describe the location distribution of equipment in the manufacturing workshop, and the equipment configuration parameters are used to describe the number of equipment at each workstation. Different equipment layout schemes will directly affect the logistics transportation path and transportation time, while different equipment quantity configurations will affect the system's processing capacity and the system's anti-disturbance redundancy capability.

[0065] Let the system design variable vector be: in For equipment layout parameters, Configure the quantity parameters for the equipment. This refers to the system's transportation time parameters.

[0066] Once the equipment layout is determined, the system automatically generates logistics transportation routes between the equipment based on their spatial coordinates and forms a transportation time matrix. The longer the transportation time between equipment, the greater the system's logistics waiting time, which can easily lead to downstream workstation starvation or upstream workstation congestion, thereby reducing the overall system production efficiency and system resilience.

[0067] Meanwhile, the number of devices configured affects the system's processing redundancy. When more parallel devices are configured at critical workstations, the system can still maintain some production capacity after a local device fails, thereby reducing the impact of disturbances on the system's production rate and improving the system's recovery capability.

[0068] After obtaining the parameter combination, a discrete-event simulation model with disturbances is run. Under the combined effects of equipment failure disturbances, maintenance disturbances, and material delay disturbances, the system's dynamic production rate changes continuously over time. The discrete-event simulation model records the instantaneous changes in the system's production rate in real time and generates a system time-production rate resilience curve.

[0069] Furthermore, the system toughness value is calculated based on the toughness curve: in, As a system resilience indicator, The instantaneous production rate of the system under disturbance conditions. The theoretical production rate of the system under undisturbed conditions. The total system simulation time. To normalize the system's production capacity, The overall disturbance intensity of the system.

[0070] The higher the system resilience index, the stronger the overall performance of the system in maintaining and restoring production capacity under complex disturbance environments.

[0071] Since system resilience calculation requires repeated running of discrete event simulation models, which has high computational cost, nonlinearity, and black-box characteristics, Bayesian optimization algorithm is used to globally optimize and solve the system parameters.

[0072] Bayesian optimization establishes a surrogate mapping model between system parameters and system resilience through Gaussian processes: in, It is a mean function. Let be the covariance function.

[0073] Gaussian process surrogate models can use existing simulation samples to predict the system resilience level in unsampled parameter regions and assess prediction uncertainty, thereby reducing the number of calculations in simulations of highly repetitive discrete events.

[0074] Furthermore, the next set of parameters to be evaluated is dynamically selected using an acquisition function: in The aim is to improve the acquisition function. The acquisition function dynamically balances high-performance local region search with global unknown region exploration, thereby improving optimization efficiency and global search capability.

[0075] During the optimization and iteration process, the system continuously updates the Gaussian process proxy model and repeatedly executes parameter generation, discrete event simulation, resilience calculation, and parameter update processes to achieve coordinated optimization of equipment layout, equipment quantity, and logistics transportation time. Ultimately, it obtains the optimal equipment layout scheme, optimal equipment configuration quantity, and optimal transportation time parameter combination that maximizes system resilience, thereby realizing the resilience optimization design of the marine engineering equipment manufacturing system under complex disturbance environments.

[0076] Step S4 specifically involves using the Sobol global sensitivity analysis method to perform global sensitivity analysis on equipment layout parameters, equipment quantity parameters, and logistics transportation time parameters, quantifying the impact of each parameter on the system resilience index. By calculating the first-order sensitivity index and the total sensitivity index of each parameter, key parameters and key workstations with a significant impact on system resilience are identified.

[0077] The system resilience model can be expressed as: in, This represents the system's toughness value. Input parameters to the system, including equipment layout parameters, equipment quantity parameters, and logistics transportation time parameters.

[0078] The Sobol method, based on the principle of variance decomposition, decomposes the total variance of system resilience into the variance contributions caused by each input parameter and the coupling effect of the parameters: in, The total variance of system resilience. For parameters Variance caused by individual effect For parameters With parameters Variance caused by coupling effect.

[0079] Further, define parameters The first-order sensitivity index is: For parameters The degree of influence of a parameter on the system's resilience when acting alone is such that the larger the first-order sensitivity index of the parameter, the more significant the direct impact of the parameter on the system's resilience.

[0080] At the same time, to account for the coupling effect between parameters, parameters are defined. The overall sensitivity index is: For parameters The overall sensitivity index, For the parameter The variance caused by other parameters.

[0081] The overall sensitivity index is used to measure parameters The overall impact on system resilience includes its own influence and the coupling effect between it and other parameters.

[0082] By calculating the first-order sensitivity index and the total sensitivity index of each parameter, the influence of system parameters is ranked, and key equipment, key workstations, and key logistics transportation links that have a significant impact on system resilience are identified, providing a basis for subsequent equipment layout optimization, equipment configuration optimization, and logistics route optimization.

[0083] Step S5 specifically involves: combining the Bayesian optimization results and the Sobol global sensitivity analysis results to comprehensively evaluate the system parameters, select the optimal parameter combination that meets the engineering constraints, and output the equipment configuration scheme, equipment spatial layout scheme, and logistics transportation time parameters that maximize system resilience.

[0084] The optimal combination of system parameters can be expressed as: L ∗ For the optimal equipment space layout parameters, N ∗ To determine the optimal number of devices, T ∗ This represents the optimal logistics transportation time parameter.

[0085] Simultaneously, the optimization results are verified for engineering feasibility to ensure that the optimized equipment layout meets workshop space constraints, safety distance constraints between equipment, and logistics transportation channel constraints, and guarantees that the number of equipment configurations meets the system's production capacity requirements. Based on the global sensitivity analysis results, key equipment, key workstations, and key logistics paths that have a significant impact on system resilience are identified, and relevant parameters are corrected and optimized in a focused manner, thereby improving the manufacturing system's resistance to disturbances and its recovery capabilities in complex disturbance environments.

[0086] Finally, the optimized equipment configuration parameters, equipment layout parameters, logistics and transportation parameters, and corresponding resilience evaluation results are stored in the system knowledge base to form reusable resilience design case data, providing data support and decision-making basis for the rapid design, parameter reconstruction, and resilience optimization of similar marine engineering equipment manufacturing production lines.

[0087] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art can make various modifications or equivalent substitutions to the present invention within its scope and spirit, and such modifications or equivalent substitutions should also be considered to fall within the scope of protection of the present invention.

Claims

1. A design method for a marine equipment resilience manufacturing system based on a Bayesian algorithm, characterized in that, Includes the following steps; Step 1: Define the system productivity as a quantitative index of the resilience of the marine engineering equipment manufacturing system, construct a discrete event simulation model of the impact of key parameters such as equipment layout, material transportation time and equipment configuration quantity on system productivity, and complete the production line model construction. Step 2: Design a multi-source disturbance injection mechanism, generate system toughness curves and calculate toughness values ​​using the discrete event simulation model constructed in Step 1; Step 3: Based on the discrete event simulation model constructed in Step 1, set the range of optimization variables, and use the Bayesian optimization algorithm based on the Gaussian process surrogate model to carry out global iterative optimization solution with the goal of maximizing system resilience; Step 4: Using the Sobol global sensitivity analysis method, the influence weight of each parameter on the system resilience is quantified, and the key bottleneck links of the manufacturing system are identified. Step 5: Output the optimal device configuration and spatial layout static parameters.

2. The method of claim 1, wherein the method is based on a Bayesian algorithm for the design of a resilient manufacturing system for maritime equipment. Step 1 specifically involves: Step 1.1: Based on the actual processing scenario of marine engineering equipment, define the equipment layout matrix, configure processing stations, logistics and transportation equipment and buffer storage areas, and complete the model building; Step 1.2: After completing the modeling of equipment processing capacity, equipment layout, and material transportation time, a system productivity calculation model is established to quantify the production capacity of the manufacturing system per unit time. theoretical conditions, the station production time; Step 1.3: Run the discrete event simulation model, dynamically adjust equipment fault parameters, maintenance sequence and system control rules, and iteratively correct and refine the details of the manufacturing system operation process.

3. The design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm according to claim 2, characterized in that, The device arrangement matrix defined in step 1.1 is specifically: device arrangement matrix The equipment arrangement matrix satisfies the following constraint conditions: each equipment can only be arranged at one position: , and at most one equipment can be arranged at each position: .

4. The method of claim 1, wherein the method is characterized by, Step 2 specifically involves: Step 2.1: Establish a multi-source disturbance mechanism that includes equipment failure disturbances and material delay disturbances, and define their respective probability distributions; Step 2.2: Define a joint perturbation injection mechanism, randomly generate event sequences according to their respective distributions on the simulation time axis, and introduce a system perturbation coupling function; Step 2.3: Run a discrete event simulation with perturbation, record the change of the instantaneous output rate Q(t) of the system over time, and plot the time-output rate curve, i.e. the resilience curve; Step 2.4: After obtaining the system time-production rate resilience curve, in order to quantitatively characterize the comprehensive performance of the marine engineering equipment manufacturing system in maintaining and restoring production capacity under disturbance conditions, the system resilience index is calculated based on the system dynamic production rate curve.

5. The design method of the Bayesian algorithm-based marine equipment resilience manufacturing system according to claim 4, characterized in that, Step 2.1 specifically involves: Define the equipment fault disturbance, and based on the probability density function of the fault occurrence, obtain the system reliability function as follows: ,in, The failure rate of device i; Define the material delay disturbance, and the actual system transportation time is: ,in, Indicates time The additional delay time generated follows a random probability distribution: P(θ) is the delay probability distribution function; This indicates the system's average transit time.

6. The design method for a robust manufacturing system for marine engineering equipment based on Bayesian algorithm according to claim 4, characterized in that, The system disturbance coupling function in step 2.2 is specifically as follows: in, The impact of equipment failure on strength, For the overall system latency intensity, To prevent the impact of maintenance on strength, Here, m represents the weighting influence coefficient; m is the total number of devices. For time The number of devices that fail at any given time. The processing capacity of equipment i; This represents the total number of transportation tasks currently in the system. For the first The delay time of each transportation task, This represents the theoretical transportation time for the corresponding task; For a moment The number of devices currently under maintenance. For equipment Maintenance duration, This represents the current simulation runtime.

7. The design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm according to claim 6, characterized in that, The instantaneous output rate Q(t) of the system in step 2.3 is specifically as follows: in, Productivity under theoretical conditions For workstation Production time.

8. The design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm according to claim 7, characterized in that, The system resilience index in step 2.4 is: .

9. The design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm according to claim 1, characterized in that, Step 4 specifically involves: using the Sobol global sensitivity analysis method to perform global sensitivity analysis on equipment layout parameters, equipment quantity parameters, and logistics transportation time parameters; and identifying key parameters and key workstations that have a significant impact on system resilience by calculating the first-order sensitivity index and total sensitivity index of each parameter.

10. The design method for a resilient manufacturing system for marine engineering equipment based on Bayesian algorithm according to claim 1, characterized in that, Step 5 specifically involves: combining the Bayesian optimization results and the Sobol global sensitivity analysis results to comprehensively evaluate the system parameters, select the optimal parameter combination that meets the engineering constraints, and output the equipment configuration scheme, equipment spatial layout scheme, and logistics transportation time parameters that maximize system resilience.

Images